Closed nethsix closed 4 years ago
Hey! Im not 100% but I think whats going on is that the image jupyter/tensorflow-notebook is clueless about the need to communikate back to the hub and say "here i am, it worked!"
~Idea 1. We change the command you exexute on the image.~ ~Idea 2. If needed, we install jupyterhub on top of the image and then does idea 1.~
Hmmm... The image is based on https://github.com/jupyter/docker-stacks... Kubespawner should update the command appropriately...
@consideRatio Yes. The image is from jupyterhub repo. However, you just reminded me of a possibly similar issue that I had with this unofficial pytorch jupyterhub image https://hub.docker.com/r/stepankuzmin/pytorch-notebook/ (https://github.com/stepankuzmin/pytorch-notebook). That image was built based on the official jupyter/scipy-notebook:latest, but I got errors trying to run it in jupyterhub (error details here), eventually I rebuilt it and it worked.
I'm closing this as it's quite old and it's a support issue. If you're still having problems please post on the Jupyter Community Forum. Thanks!
Platform: google cloud Jupyterhub helm chart: v0.6.0-d359d98
I successfully setup and run zero-to-jupyterhub-k8s. However a day or so, I will no longer be able to login.
Initially I thought this could be resource issue, so I upgraded my k8s nodes to 'n1-highmem-2', which is:
Upon checking further, from the logs I show below, it seems that the hub is complaining about the jupyter pod not spawning quickly enough but the logs on the jupyter pod shows that it is already running.
I thought I could resolve this by setting http_timeout, start_timeout of the hub (see config.yaml), but it has no effect whatsoever.
I deployed using helm as follows:
helm install jupyterhub/jupyterhub --version=v0.6.0-d359d98 --name=fastowl-tf-timeout-jupyter --namespace=fastowl-tf-timeout-space -f config/k8s-config-github-tf-timeout.yml --debug --timeout=99999
Please help! Thank you.
Contents of
config.yaml
Hub logs:
Jupyter node logs: